Information processing apparatus, information processing method, and program

Abstract

To predict power demand on the basis of predicted behavior of a consumer, and to predict information on power provided from a distributed energy source.SOLUTION: The information processing device 1 includes a departure prediction unit 10 that predicts departure of a consumer in a predetermined region, and a new extension prediction unit 15 that predicts new extension of a consumer in the predetermined region, and predicts power demand in the predetermined region using information regarding the predicted departure of the consumer and information regarding the predicted new extension of the consumer. An information processing device 1 includes an introduction prediction unit 30 that predicts introduction information on introduction of a distributed energy source of a consumer in a predetermined area, and predicts information on power provided from the distributed energy source in the predetermined area by using the predicted introduction information.SELECTED DRAWING: Figure 7

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and a program for predicting power demand and predicting information related to power provided by distributed energy sources. [Background technology]

[0002] Electricity demand forecasts have been made for some time, and attempts have been made to improve the accuracy of such forecasts. For example, Patent Document 1 proposes generating, for each of a plurality of customer groups, a demand forecasting model that represents the relationship between fluctuation factor data and electricity demand data based on a plurality of training data accumulated in a database, obtaining fluctuation factor forecast data for each of the plurality of customer groups, calculating electricity demand forecast data for each of the plurality of customer groups based on the fluctuation factor forecast data and the demand forecasting model, and calculating total electricity demand forecast data for an electricity seller based on the electricity demand forecast data calculated for each of the plurality of customer groups. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-084328 Summary of the Invention [Problem to be solved by the invention]

[0004] Although the invention disclosed in Patent Document 1 aims to improve the accuracy of power demand forecasts, it does not forecast power demand based on the predicted behavior of consumers, nor does it anticipate predicting information about power, such as power and current, provided by distributed energy sources based on the predicted behavior of consumers.

[0005] The present invention provides an information processing device, an information processing method, and a program for predicting power demand based on predicted consumer behavior and predicting information related to power provided by distributed energy sources. [Means for solving the problem]

[0006] [Concept 1] The information processing device according to the present invention comprises: a churn prediction unit that predicts churn of consumers in a predetermined area; a new installation / expansion prediction unit that predicts new installation / expansion of customers in a predetermined area; Equipped with The power demand in a predetermined area may be predicted using information regarding predicted customer departures and information regarding predicted new or additional customer installations.

[0007] [Concept 2] The information processing device according to Concept 1 is The system may further include a power demand and supply prediction unit that predicts the power flow in the power distribution facilities installed in the predetermined area using information about predicted customer departures and information about new or additional customer installations.

[0008] [Concept 3] The information processing device according to concept 1 or 2 is a dividing unit that divides the predetermined area into a plurality of cells, The demand forecasting unit forecasts customer withdrawal for each cell, The new installation / expansion prediction unit may predict new installation / expansion of customers for each cell.

[0009] [Concept 4] In an information processing device according to any one of concepts 1 to 3, The demand forecasting unit predicts customer departures using land use classification, The new installation / expansion prediction unit may predict new installation / expansion of consumers using land use classification.

[0010] [Concept 5] The information processing device according to Concept 4 is The land use classification prediction unit may be provided to predict the land use classification using past performance.

[0011] [Concept 6] In an information processing device according to any one of concepts 1 to 5, The demand forecasting unit predicts customer churn using regional characteristics, The new installation / expansion prediction unit may predict new installation / expansion of customers using regional characteristics.

[0012] [Concept 7] In the information processing device according to concept 3, The demand forecasting unit predicts customer churn for each cell using land use classification and regional characteristics, The new installation / expansion prediction unit may predict new installation / expansion of customers for each cell using land use classification and regional characteristics.

[0013] [Concept 8] In an information processing device according to any one of concepts 1 to 7, The demand forecasting unit predicts customer churn using a simulation method, The new installation / expansion prediction unit may predict new installation / expansion of customers using a simulation method.

[0014] [Concept 9] The information processing device according to Concept 3 is a classification prediction unit that predicts a land use classification for each cell using a cellular automaton; the demand forecasting unit predicts customer churn for each cell by performing a multi-agent simulation using land use classification for each cell; The new installation / expansion prediction unit may predict new installation / expansion of consumers for each cell by performing a multi-agent simulation using land use classification for each cell.

[0015] [Concept 10] The information processing device according to the present invention comprises: an introduction prediction unit that predicts introduction information regarding introduction of distributed energy sources by consumers in a predetermined area; The predicted introduction information may be used to predict information regarding the power provided by distributed energy sources in a given area.

[0016] [Concept 11] 11. The information processing device according to claim 10, The power supply and demand forecasting unit may be configured to forecast a power flow in a power distribution facility installed in the predetermined area using the information on the introduction of distributed energy sources.

[0017] [Concept 12] In the information processing device according to Concept 10 or 11, The introduction prediction unit may predict information regarding the introduction of distributed energy resources in association with existing customers or regions.

[0018] [Concept 13] 13. An information processing device according to any one of claims 10 to 12, A division unit is provided for dividing a predetermined area into a plurality of cells, The introduction prediction unit may predict the introduction information of the distributed energy source for each cell.

[0019] [Concept 14] 10. An information processing device according to any one of concepts 1 to 9, an introduction prediction unit that predicts introduction information regarding the introduction of distributed energy sources of consumers in the predetermined area; an electric power demand and supply prediction unit that predicts electric power flows in electric power distribution facilities installed in the specified area using information predicting the churn of consumers in the specified area predicted by the churn prediction unit, information regarding new or additional consumer installations, and the introduction information of distributed energy sources; may also be provided.

[0020] [Concept 15] The information processing method according to the present invention comprises: a step of predicting churn of consumers in a predetermined area by a churn prediction unit; a step of predicting new installations or additions of customers in a predetermined area by a new installation or addition prediction unit; Equipped with The power demand in a predetermined area may be predicted using information regarding predicted customer departures and information regarding predicted new or additional customer installations.

[0021] [Concept 16] The program according to the present invention comprises: A program to be installed on an information processing device, On the information processing device on which the program is installed, A function to predict customer churn in a specified area; A function to predict new or additional installations of customers in a specified area; A function for predicting the demand for electricity in a predetermined area using information on predicted customer departures and information on predicted new or additional customer installations; may be realized.

[0022] [Concept 17] The information processing method according to the present invention comprises: a step of predicting introduction information regarding introduction of distributed energy sources of consumers in a predetermined area by an introduction prediction unit; using the predicted introduction information to predict information regarding the power provided by distributed energy sources in a predetermined area; may also be provided.

[0023] [Concept 18] The program according to the present invention comprises: A program to be installed on an information processing device, On the information processing device on which the program is installed, A function for predicting the introduction information regarding the introduction of distributed energy sources by consumers in a specified area; a function for predicting information about the power provided by distributed energy sources in a predetermined area using the predicted introduction information; may be realized. [Effects of the Invention]

[0024] According to the present invention, it is possible to predict power demand based on predicted consumer behavior, and to predict information related to power such as power and current provided from distributed energy sources. [Brief explanation of the drawings]

[0025] [Figure 1] FIG. 2 is a diagram for explaining an example of a manner for obtaining land use classification information for a target area in an embodiment of the present invention. [Figure 2] FIG. 10 is a diagram for explaining an example of a mode for calculating a departure probability and a new expansion probability in an embodiment of the present invention. [Figure 3] FIG. 1 is a diagram for explaining an example of a mode in which the probability of each customer leaving the grid and the probability of new installation or expansion in each grid (cell) are calculated using a customer leaving prediction model and a customer new installation or expansion prediction model in an embodiment of the present invention. [Figure 4] FIG. 10 is a diagram showing a mode in which it is determined for each cell whether an existing customer will leave, and whether a customer will be newly added or expanded in each cell, in an embodiment of the present invention. [Figure 5] FIG. 10 is a diagram showing an aspect in which an introduction prediction unit predicts, for each cell, whether to purchase and introduce a predetermined distributed energy source by using a multi-agent simulation (MAS) in an embodiment of the present invention. [Figure 6] FIG. 10 is a diagram showing an aspect in which an introduction prediction unit predicts, for each cell, whether to newly install and introduce a predetermined distributed energy source using a multi-agent simulation (MAS) in an embodiment of the present invention. [Figure 7] 1 is a schematic diagram showing a configuration of an information processing device according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0026] Embodiment Hereinafter, an embodiment of an information processing device and an information processing method according to the present invention will be described. In this embodiment, a program that can be installed in a computer such as a personal computer to enable the computer to execute the information processing method according to the embodiment, and a recording medium on which the program is recorded, are also provided. In this embodiment, "or" is a concept that includes "and," and "A or B" is a concept that includes only A, only B, or both A and B.

[0027] The information processing device of this embodiment may be installed anywhere, may be a server, or may be used in a cloud environment. The information processing device of this embodiment may be composed of one device or multiple devices. Furthermore, when an information processing device is composed of multiple devices, the devices do not need to be installed in the same space such as the same room, but may be installed in different rooms, different buildings, different regions, etc. Furthermore, when an information processing device is composed of multiple devices, some of the devices may be owned and / or managed by one institution, and the rest may be owned and / or managed by another institution.

[0028] The information processing device of this embodiment is generated, for example, by installing a program. This program may be distributed by email, may be available by accessing a predetermined URL and logging in, or may be recorded on a recording medium. The program of this embodiment is used to generate the information processing device described below, and the recording medium of this embodiment is used to record the program. Furthermore, the information processing method of this embodiment is implemented by an information processing device in which the program is installed. The information processing device may execute the information processing method of this embodiment by executing an application installed on the information processing device.

[0029] As shown in FIG. 7 , the information processing device 1 of this embodiment may include a churn prediction unit 10 that predicts customer churn in a predetermined area and a new / addition prediction unit 15 that predicts new or additional installations of customers in the predetermined area. In this embodiment, "new or additional installations" refers to a situation in which a customer moves, resulting in an increase in contract capacity, due to new installations, additions, capacity changes, etc. Furthermore, "churn" in this embodiment refers to a situation in which a customer moves, resulting in a decrease in contract capacity, due to closure, removal, capacity changes, etc. The information processing device 1 may include a power supply and demand prediction unit 20 that predicts power demand in a predetermined area using information on predicted customer churn and information on predicted new or additional installations of customers. The power supply and demand prediction unit 20 may predict power flow in a predetermined power distribution facility, or may predict the current, voltage, or power flowing through the power distribution facility, using information on customer churn and new or additional installations. The predetermined power distribution facility may be a main power distribution facility installed in the predetermined area, or each of the power distribution facilities installed in the predetermined area. Therefore, as an example, the power demand and supply prediction unit 20 may predict the power flow in each of the power distribution facilities installed in a predetermined area using information about customer departures and new installations. The departure prediction unit 10 may predict the probability of customer departures, and the new installation prediction unit 15 may predict the probability of customer new installations. The information processing device 1 of this embodiment may perform an overall prediction for customer departures or new installations in a predetermined area, and then perform a local prediction for customer departures or new installations in the predetermined area so that the prediction matches the overall prediction for the predetermined area.

[0030] An adjustment unit 60 may be provided that adjusts the numerical value for a specific region so that it matches the numerical value for the entire region (also referred to as "whole system") in the specific region. The adjustment unit 60 may adjust the total cell defect probability so that it matches the overall defect probability in the specific region. Similarly, the adjustment unit 60 may adjust the total new expansion probability for cells so that it matches the overall new expansion probability in the specific region.

[0031] The total system customer number forecast may predict fluctuations in the number of customers in a base year and determine the potential increase in the number of customers across the entire service area (total system). The local forecast may perform a regional distribution forecast, which may identify areas where new customers will be created and allocate the total number of customers to smaller areas. The local forecast may also use a demand curve estimate, a peak demand estimate, or a demand volume estimate to set the time-of-day change in demand per customer. The regional distribution forecast and the demand curve estimate, peak demand estimate, or demand volume estimate may then be used to multiply the "increase in the number of customers" by the "demand per customer."

[0032] Generally speaking, the trend method can be used when the parameter is large, but when the trend method is used when the parameter is small, long-term patterns and trends tend to disappear, resulting in reduced prediction accuracy. For this reason, when the parameter is small, it is more beneficial to use the simulation method rather than the trend method. When the parameter is small in a specific region as described in this embodiment, it is more beneficial to use the simulation method rather than the trend method. On the other hand, the trend method may be used for overall (system-wide) predictions in a specific region where the parameter is large.

[0033] Examples of trend methods include curve regression analysis, multivariate regression analysis, and template matching. In curve regression analysis, actual peak demand may be fitted to a growth curve (an S-shaped curve such as a Gompertz curve) to predict future peak demand. In multivariate regression analysis, actual peak demand may be regressed with external statistics such as economic indicators and population trends to predict future peak demand. In template matching, actual peak demand in other areas may be searched for that has a similar shape to the actual peak demand, and the trends in the matched areas may be applied to future predictions. As the trend method, any technology that is publicly known at the time of filing of this application may be used as appropriate.

[0034] The simulation method may be a cellular automata (CA), a multi-agent simulation (MAS), or a combination of both.

[0035] A cellular automata is an "automatic machine" in which a large number of cells arranged in a grid in space change their state over time as they interact with neighboring cells. It is characterized by the fact that the state of each cell takes on discrete values ​​(such as "0," "1," "2," etc.), that each cell interacts locally with cells within a certain range (neighborhood), and that the state of a cell in the next period is determined deterministically from the current states of itself and its neighboring cells. A cellular automata (CA) is a discrete computational model based on cells and simple rules.

[0036] In multi-agent simulation (MAS), an "agent" is an entity that recognizes its surroundings and, based on that, acts autonomously under certain rules. An agent can be said to be an entity that behaves autonomously by simulating the behavior of humans or living things as rules. Multi-agent simulation is a simulation (virtual experiment) in which multiple (multiple) agents (people, living things, etc.) simultaneously execute their tasks based on their own rules, while interfering with (interacting with) each other. In multi-agent simulation for demand forecasting, consumers who sign electricity contracts can be considered agents.

[0037] Simulation methods including cellular automata (CA) and multi-agent simulation (MAS) can be appropriately made use of techniques that are publicly known at the time of filing of this application.

[0038] As described above, the information processing device 1 of this embodiment may perform a forecast of the total number of consumers using information on consumer departures and new installations, and then perform a local forecast using information on consumer departures and new installations in a specified area. For the forecast of the total number of consumers, a trend method may be used because the parameter is large, and for the local forecast, a simulation method may be used.

[0039] The target area for local prediction may be specified by a user through input from a terminal 100 such as a personal computer, tablet, or smartphone, or may be selected via the terminal 100 from a plurality of supply areas pre-stored in the storage unit 80. For example, when a point on a map is selected via the terminal 100, a target area including the point may be selected. The information processing device 1 of this embodiment is capable of communicating with the terminal 100 via a communication unit 5 having the functions of an output unit that outputs information and a receiving unit that receives information.

[0040] A dividing unit 50 may be provided that divides a predetermined area into a plurality of cells (meshes). In Figures 1 and 2, "cells" are expressed as "meshes." The departure prediction unit 10 may predict customer departures for each cell. The new installation / addition prediction unit 15 may predict new installation / addition of customers for each cell. In this case, the departure prediction unit 10 may predict the probability of customer departures for each cell. The new installation / addition prediction unit 15 may predict the probability of new installation / addition of customers for each cell. Cells are not limited to squares, and various shapes such as rectangles, trapezoids, rhombuses, and parallelograms may be used. Furthermore, predictions may be made for each predetermined location using latitude and longitude, etc., without using cells.

[0041] The demand forecasting unit 10 may use cellular automata (CA) to predict the land use classification of a predetermined area and may use multi-agent simulation (MAS) to predict customer departures for each cell. Similarly, the new installation / expansion forecasting unit 15 may use cellular automata (CA) to predict the land use classification of a predetermined area and may use multi-agent simulation (MAS) to predict customer installation / expansion for each cell.

[0042] The demand forecasting unit 10 may predict customer departures using land use classifications such as residential areas, roads, parks, and forests, and the new installation / expansion forecasting unit 15 may predict new installation / expansion of customers using land use classifications.

[0043] A classification prediction unit 70 may be provided to predict land use classification. The classification prediction unit 70 may predict land use classification using past performance data. Regarding land use classification performance data, as shown in FIG. 1, a land use classification model may be generated using satellite image data for a predetermined period, such as by year, and land use classification performance data for a predetermined period, such as by year. Using satellite imagery makes it possible to determine whether a target cell is mostly residential or mostly forested. Therefore, if the target cell is mostly residential, it is determined as a "residential area," and if the target cell is mostly forested, it is determined as a "forest." The largest land classification in a cell may be used to classify the cell. For example, if a cell is 40% "residential area," 35% "road," and 25% "park," the land use prediction model may be generated so that the cell is determined as a "residential area." Alternatively, if a cell is 50% "high-rise apartment buildings," 40% "detached houses," and 10% "road," the land use prediction model may be generated so that the cell is determined as a "high-rise apartment building." For example, a model for determining land use classification from a satellite image may be generated by machine learning using satellite image data for each cell and actual land use classification. Then, a land use classification model may be generated by machine learning using satellite image data for each predetermined period, such as by year, and actual land use classification results for each predetermined period, such as by year, as training data. The size of the cells may be set to a predetermined size, and the satellite image may be divided into cells of a fixed size, such as 100 m x 100 m to 300 m x 300 m.

[0044] The land use classification may be predicted for each cell divided by the dividing unit 50. In this case, for example, the land use classification of the own cell and neighboring cells in year n and the land use classification of the own cell in year n+1 may be used to predict the future land use classification of the own cell. In this case, the prediction may be performed using a Markov chain or machine learning. When machine learning is used, a land use classification model may be generated using the land use classification of the own cell and neighboring cells in year m and the land use classification of the own cell in year m+1 over multiple years as training data, and the land use classification of the own cell for a target year (2050 in FIG. 1 ) may be predicted using the land use classification model.

[0045] When the model described in this embodiment is generated using machine learning, the model may be generated by the generation unit 40, or may be generated separately and stored in the storage unit 80. The explanatory variables (elements) to be used and their coefficients (weights) may be determined using machine learning techniques. Various models can be used in machine learning, and for example, a tree model, a random forest model, or the like may be used.

[0046] The demand forecasting unit 10 may predict customer departures using distances from features. The new installation / expansion forecasting unit 15 may predict customer new installation / expansion using distances from features. As the distance from a feature, regional characteristics based on location information of the customer may be used, and for example, the distance from a predetermined feature, such as the distance from the center of a city, may be used.

[0047] As an example, as shown in Figure 2, the churn probability in a target area may be calculated using the land use classification (see Figure 1) obtained by using a land use classification model and regional characteristics based on location information of consumers as feature quantities. Although "other indicators" are shown in Figure 2, external indicators such as GDP may also be used as "other indicators." Furthermore, the new construction probability in the target area may be calculated using the land use classification, distance from features, and other indicators as feature quantities. The churn probability may be calculated for each cell, or the new construction probability may be calculated for each cell.

[0048] It is common for there to be a period of time between a change in the land use classification for each cell and consumers moving into that cell, but by predicting the land use classification for each cell and then simulating whether consumers will move into that cell, it is possible to make predictions that reflect this fact.

[0049] The number of customer churn for each cell may also be obtained by multiplying the number of existing customers by the churn probability. Reasons for churn include natural attrition of households and relocation due to relocation within or outside the area. Meanwhile, reasons for new construction and expansion include natural increase of households and relocation due to relocation within or from the area. The probability of churn and the probability of new construction and expansion may be calculated using information on the master development plan area as "other indicators," in addition to the regional characteristics based on land use classification and customer location information. Because information on land use classification, regional characteristics based on customer location information, and master development plan area tends to change over time, it is preferable to update this information as appropriate. By adopting this approach, predictions regarding churn and new construction and expansion can be made using the latest data.

[0050] The generation unit 40 may use, as learning data, information on past land use classifications and distances from features, other indicators such as the master development plan area, and information on past customer churn (for example, the churn probability and whether each customer churns or remains), to generate a customer churn prediction model according to a decision-making rule, and store the model in the storage unit 80 (see FIG. 3). Similarly, the generation unit 40 may use, as learning data, information on past land use classifications and distances from features, other indicators such as the master development plan area, and information on past customer new installations or expansions (for example, the new installation probability and whether each customer installed or did not install a new installation), to generate a customer new installation or expansion prediction model according to a decision-making rule, and store the model in the storage unit 80 (see FIG. 3). A deep neural network may be used in the decision-making rule.

[0051] The customer churn probability for each cell may be calculated using a customer churn prediction model, and the number of customer churns for each cell may be calculated by multiplying the churn probability by the number of customers in the cell. If it is difficult to obtain the number of customers for each cell as data, the number of customer churns for each cell may be calculated by multiplying the churn probability for each cell by the value obtained by dividing the total predicted number of customers leaving the target area by the number of cells. Alternatively, the customer new addition / installation probability for each cell may be calculated using a customer new addition / installation prediction model, and the number of customer new addition / installation for each cell may be calculated by multiplying the calculated new addition / installation probability by the value obtained by dividing the total predicted number of customers newly entering the target area by the number of cells.

[0052] The total predicted number of consumers leaving the target area and the total predicted number of consumers entering the target area may be predicted from past statistical data, or for simplicity, the most recent statistical data (e.g., last year) may be used. The total increase or decrease in the number of consumers in the target area may be calculated from the statistical data, and the number of new installations in the target area may be calculated by adding the total number of consumers to the number of departures in the target area. The calculation unit 110 may perform the process of calculating the total increase or decrease in the number of consumers in the target area and the process of calculating the number of new installations in the target area from the statistical data. Once the number of departures and the number of new installations in the target area are derived in this way, the calculation unit 110 may calculate the number of departures or the departure probability for each cell in a manner consistent with the total number of departures, and similarly, the calculation unit 110 may calculate the number of new installations or the probability of new installations for each cell in a manner consistent with the total number of new installations. The calculation unit 110 may be provided in addition to the adjustment unit 60 described above, or only one of the adjustment unit 60 and the calculation unit 110 may be provided.

[0053] A multi-agent simulation (MAS) may be adopted to simulate the behavior of existing customers leaving and new customers choosing to install or install new facilities using a customer leaving probability prediction model such as a customer leaving probability prediction model and a customer new facility addition probability prediction model. In this case, as shown in Figure 2, it may be possible to predict whether an agent consisting of existing customers will leave for each cell. It may also be possible to predict whether an agent will install or install new facilities for each cell.

[0054] In the embodiment shown in Fig. 4, it is determined for each cell whether an existing customer (agent) will leave, and if they will leave, it is shown as "Leave (1)", and if they will not leave, it is shown as "Maintain (0)". It is also determined for each cell whether a customer will install a new or additional facility, and if a customer (agent) will not install a new or additional facility, it is shown as "No new or additional facility (0)", and if they will install a new or additional facility, it is shown as "New or additional facility (1)".

[0055] An introduction prediction unit 30 may be provided to predict introduction information of distributed energy sources for consumers in a specified area. The power demand and supply prediction unit 20 of the information processing device 1 may predict information about power, such as the power and current provided by distributed energy sources in a specified area, using the introduction information about the predicted introduction of distributed energy sources. The power demand and supply prediction unit 20 may predict the power flow in a specified power distribution facility using the introduction information about distributed energy sources, or may predict the current, voltage, or power flowing through the specified power distribution facility. The specified power distribution facility may be a main power distribution facility installed in the specified area, or each of the power distribution facilities installed in the specified area. When using multi-agent simulation for new installations, expansions, and removals, as described above, it is possible to determine agents (consumers) for each cell, so one or more agents in each cell may determine whether to purchase and introduce a specified distributed energy source. The introduction prediction unit 30 may predict the probability of purchasing a distributed energy source or the probability of introducing a distributed energy source as introduction information. Examples of distributed energy sources include electric vehicles (EVs), heat pumps, storage batteries, and solar cells.

[0056] The introduction prediction unit 30 may predict information related to the introduction of distributed energy resources, such as the probability of introduction, in association with a predetermined existing consumer or a predetermined region. The predetermined existing consumer or the predetermined region may be a consumer or a region designated by a user. The predetermined region may be a region with no existing consumer or a region not associated with an existing consumer.

[0057] The introduction prediction unit 30 may use a multi-agent simulation to determine whether an agent will purchase and introduce a predetermined distributed energy source for each cell. Similarly, the introduction prediction unit 30 may use a multi-agent simulation to determine whether an agent will newly install and introduce a predetermined distributed energy source for each cell. The introduction prediction unit 30 may determine whether an agent will purchase and introduce, for each cell, various types of distributed energy sources, such as EVs, heat pumps, storage batteries, and solar cells (on-site). The introduction prediction unit 30 may determine whether an agent will newly install and introduce, for each cell, a distributed energy source, such as solar cells (outdoor installation). Furthermore, the introduction prediction unit 30 may determine whether a predetermined distributed energy source will be purchased and introduced for each consumer. Similarly, the introduction prediction unit 30 may determine whether an agent will newly install and introduce a predetermined distributed energy source for each consumer.

[0058] The introduction prediction unit 30 may predict whether or not to introduce each type of distributed energy resource. For existing consumers, the introduction prediction unit 30 may use consumer attributes such as income information and household information, and information on existing distributed energy resources already owned (e.g., contract capacity, purchase date, etc.) to determine whether or not to purchase and introduce each type of distributed energy resource. This is because, for example, consumers with ample resources tend to purchase multiple distributed energy resources or replace them frequently. Information on consumer attributes such as consumer income and family structure may be obtained by using average values ​​presented in statistical data. Information on existing distributed energy resources may be obtained by using information already held by users such as electric power companies.

[0059] Because the decision to newly introduce a distributed energy resource is tied to the land, a cell division approach may be adopted. On the other hand, because the decision to purchase a distributed energy resource by an existing consumer is tied to the consumer, a cell division approach may not be adopted. That is, a simulation may be performed for each consumer to determine whether to purchase the target distributed energy resource (see "Introduction (1)" or "No Introduction (0)" in FIG. 5). However, because dividing into cells and using the average value of various information of existing consumers in each cell can reduce the number of simulations, a simulation (multi-agent simulation) may also be performed for existing consumers to determine whether to purchase a distributed energy resource. In this case, because there may be variation in the purchasing decisions of consumers within a cell, a purchase probability may be calculated. More specifically, for each type of distributed energy resource, the feature values ​​of existing consumers or the average value of the feature values ​​of consumers in each cell may be applied to a DER purchase prediction model (a type of DER introduction prediction model) such as a DER purchase probability prediction model to predict whether to purchase or maintain the distributed energy resource without purchasing it.

[0060] Furthermore, when predicting whether existing consumers will purchase or maintain a distributed energy resource by applying the characteristics of existing consumers to a DER purchase prediction model, the simulation results for existing consumers in the cell may be aggregated for each cell, and the purchase probability may be calculated for each cell. As an example, the introduction prediction unit 30 may predict whether or not to introduce each type of distributed energy resource for each agent by applying the characteristics of the land in the cell to a DER introduction prediction model, such as a DER introduction probability prediction model. In this case, since the simulation results may differ for each agent, the purchase probability may be calculated for each cell using the total number of agents as a parameter.

[0061] The decision-making rules in the simulation (multi-agent simulation) in this embodiment may be achieved by an evolutionary algorithm (genetic algorithm), reinforcement learning (Q-learning or deep Q-learning), a decision-making model, or the like. These may also be probabilistic models. This decision-making rule may be used not only in the customer churn prediction model and the customer new installation prediction model described above, but also in a DER introduction prediction model.

[0062] In addition to the introduction, a disposal prediction unit 35 may be provided that predicts information regarding disposal, such as the probability of disposal of distributed energy resources, for existing consumers. In this case, both the disposal of existing distributed energy resources and the introduction of distributed energy resources are predicted for existing consumers. Discarded distributed energy resources may be specific distributed energy resources that are predicted to be replaced by EVs, etc.

[0063] The introduction prediction unit 30 may predict, for each cell, introduction information regarding whether or not a distributed energy source will be introduced. For example, in the case of EVs, the introduction prediction unit 30 may use a multi-agent simulation (MAS) to determine, for each cell, whether existing consumers will purchase and introduce a predetermined distributed energy source (e.g., EV) (see FIG. 5). The introduction prediction unit 30 may also use a multi-agent simulation (MAS) to determine, for each cell, whether or not a predetermined distributed energy source (e.g., PV) will be newly installed and introduced (see FIG. 6).

[0064] The generation unit 40 may use, as learning data, information such as the type of distributed energy resource to be introduced, consumer attributes such as income information and family structure, information on existing distributed energy resources already owned, and information on whether consumers in the past have actually purchased and introduced the target distributed energy resource, to generate a DER introduction prediction model such as the purchase decision-making probability prediction model shown in FIG. 5 through machine learning, and store the model in the storage unit 80. Alternatively, the generation unit 40 may use, as learning data, information such as the type of distributed energy resource to be introduced, land use classification, and information on whether consumers in the past have actually newly installed and introduced the target distributed energy resource, to generate a DER introduction prediction model such as the new installation probability prediction model shown in FIG. 6 through machine learning, and store the model in the storage unit 80. A deep neural network may be used in the machine learning.

[0065] The power supply and demand forecasting unit 20 may predict the power absorbed (power consumed) from power distribution facilities installed in a predetermined area using information obtained by the withdrawal forecasting unit 10 and the new installation forecasting unit 15. The power supply and demand forecasting unit 20 may predict the power supplied to power distribution facilities installed in a predetermined area using information obtained by the introduction forecasting unit 30. In this way, the power (e.g., power flow) at the predetermined power distribution facility may be predicted. As described above, the predetermined power distribution facility may be a main power distribution facility installed in the predetermined area, or each of the power distribution facilities installed in the predetermined area.

[0066] In addition, the power demand and supply prediction unit 20 may predict the power (e.g., power flow) not only for the power distribution facilities installed in the target area (specified area), but also for the power distribution facilities installed in non-target areas (outside the specified area).

[0067] The communication unit 5, demand forecasting unit 10, new installation / expansion forecasting unit 15, power demand / supply forecasting unit 20, installation forecasting unit 30, generation unit 40, division unit 50, classification / prediction unit 70, calculation unit 110, etc. may be realized by a single unit (control unit) or by different units. The functions of multiple "units" may be integrated; for example, the functions of the demand forecasting unit 10 and the new installation / expansion forecasting unit 15 may be realized by a single unit. Furthermore, the communication unit 5, demand forecasting unit 10, new installation / expansion forecasting unit 15, power demand / supply forecasting unit 20, installation forecasting unit 30, generation unit 40, division unit 50, classification / prediction unit 70, calculation unit 110, etc. may be realized by a circuit configuration or by a processor executing a program.

[0068] The above description of the embodiment and the disclosure of the drawings are merely examples for explaining the invention described in the claims, and the invention described in the claims is not limited by the above description of the embodiment or the disclosure of the drawings. [Explanation of symbols]

[0069] 1. Information processing equipment 10 Demand Forecasting Department 15 New Construction Forecasting Department 20 Electricity Demand and Supply Forecasting Department 30 Introduction Forecasting Department 40 Generation part 50 divisions 70 Classification and Prediction Section 80 Storage section

Claims

1. a churn prediction unit that predicts churn of consumers in a predetermined area; a new installation / expansion prediction unit that predicts new installation / expansion of customers in a predetermined area; Equipped with An information processing device that predicts power demand in a specified area using information regarding predicted customer departures and information regarding predicted new or additional customer installations.

2. The information processing device according to claim 1 , further comprising: a power demand and supply prediction unit that predicts a power flow in a power distribution facility installed in the specified area using information about predicted customer withdrawals and information about new or additional customer installations.

3. a dividing unit that divides the predetermined area into a plurality of cells, The demand forecasting unit forecasts customer withdrawal for each cell, The information processing device according to claim 1 , wherein the new installation / expansion prediction unit predicts new installation / expansion of a customer for each cell.

4. The demand forecasting unit predicts customer departures using land use classification, The information processing device according to claim 1 , wherein the new construction or expansion prediction unit predicts new construction or expansion of customers using land use classification.

5. The information processing device according to claim 4 , further comprising a classification prediction unit that predicts the land use classification using past performance.

6. The demand forecasting unit predicts customer churn using regional characteristics, The information processing device according to claim 1 , wherein the new installation / expansion prediction unit predicts new installation / expansion of customers using regional characteristics.

7. The demand forecasting unit predicts customer churn for each cell using land use classification and regional characteristics, The information processing device according to claim 3 , wherein the new installation / expansion prediction unit predicts new installation / expansion of customers for each cell using land use classification and regional characteristics.

8. The demand forecasting unit predicts customer churn using a simulation method, The information processing device according to claim 1 , wherein the new installation / expansion prediction unit predicts new installation / expansion of customers using a simulation method.

9. a classification prediction unit that predicts a land use classification for each cell using a cellular automaton; the demand forecasting unit predicts customer departures for each cell by performing a multi-agent simulation using land use classification for each cell; The information processing device according to claim 3 , wherein the new installation / expansion prediction unit predicts new installation / expansion of consumers for each cell by performing a multi-agent simulation using land use classification for each cell.

10. an introduction prediction unit that predicts introduction information regarding introduction of distributed energy sources by consumers in a predetermined area; An information processing device that uses the predicted introduction information to predict information regarding the power provided by distributed energy sources in a specified area.

11. The information processing device according to claim 10 , further comprising: an electric power demand and supply prediction unit that predicts an electric power flow in a power distribution facility installed in the predetermined area using the introduction information of the distributed energy source.

12. The information processing device according to claim 10 or 11, wherein the introduction prediction unit predicts information regarding the introduction of distributed energy sources in association with existing customers or regions.

13. A division unit is provided for dividing a predetermined area into a plurality of cells, The information processing device according to claim 10 , wherein the introduction prediction unit predicts the introduction information of the distributed energy source for each cell.

14. an introduction prediction unit that predicts introduction information regarding the introduction of distributed energy sources of consumers in the predetermined area; an electric power demand and supply prediction unit that predicts electric power flows in electric power distribution facilities installed in the specified area using information predicting the churn of consumers in the specified area predicted by the churn prediction unit, information regarding new or additional consumer installations, and the introduction information of distributed energy sources; The information processing device according to claim 1 , comprising:

15. a step of predicting churn of consumers in a predetermined area by a churn prediction unit; a step of predicting new installations or additions of customers in a predetermined area by a new installation or addition prediction unit; Equipped with An information processing method for forecasting electricity demand in a specified area using information regarding predicted customer departures and predicted new or additional customer installations.

16. A program to be installed on an information processing device, On the information processing device on which the program is installed, A function to predict customer churn in a specified area; A function to predict new or additional installations of customers in a specified area; A function for predicting the demand for electricity in a predetermined area using information on predicted customer departures and information on predicted new or additional customer installations; A program that makes this happen.

17. a step of predicting introduction information regarding introduction of distributed energy sources of consumers in a predetermined area by an introduction prediction unit; using the predicted introduction information to predict information regarding the power provided by distributed energy sources in a predetermined area; An information processing method comprising:

18. A program to be installed on an information processing device, On the information processing device on which the program is installed, A function for predicting the introduction information regarding the introduction of distributed energy sources by consumers in a specified area; a function for predicting information about the power provided by distributed energy sources in a predetermined area using the predicted introduction information; A program that makes this happen.

Images